more in the data structure survey

Data-structure-survey.md

@@ -55,6 +55,10 @@ Fixed-size array stack
 
   * `isc/netmgr/netmgr`
 
+## notes
+
+Locked.
+
 
 # `isc/heap`
 
@@ -68,7 +72,7 @@ Priority queue
 
 ## notes
 
-No locking / single-threaded.
+Single-threaded.
 
 
 # `isc/ht`
@@ -84,12 +88,26 @@ Hash table
 
 ## notes
 
-No locking / single-threaded.
+Single-threaded.
 
 Uses chaining, so it is heavy on indirection and heavy on allocation.
 Modern hash tables use open addressing.
 
 
+# `isc/pool`
+
+A mechanism for sharing a small pool of fungible objects among a large
+number of objects that depend on them.
+
+## used by
+
+  * `dns/zone` for sharing memory contexts
+
+## notes
+
+Single-threaded.
+
+
 # `isc/radix`
 
 Radix tree for IP address prefix matching
@@ -101,6 +119,8 @@ Radix tree for IP address prefix matching
 
 ## notes
 
+Single-threaded.
+
 Slightly unusual because it is first-match rather than best-match.
 
 Tests one bit at a time so there's an opportunity to improve with
@@ -108,3 +128,25 @@ Tests one bit at a time so there's an opportunity to improve with
 
 The nodes appear to be larger than necessary; is this a problem?
 How big do ACLs get?
+
+Uses reserved identifiers for static functions, which is undefined
+behaviour.
+
+
+# `isc/taskpool`
+
+Like `isc/pool`, but specialized for tasks
+
+## used by
+
+  * `dns/zone` for sharing tasks
+
+## notes
+
+Single-threaded.
+
+Should probably be replaced by the more generic `isc/pool` (it looks
+like that was the plan when `isc/pool` was added).
+
+There appears to be a bug in `isc_taskpool_expand()` when it fails to
+create a new task: the existing taskpool will be left empty.